| At present, with the rapid development of web 2.0, online social networks are growing popularity. Public opinion thus is integrated into online and offline networks gradually, and the regional characteristics of network public opinion is becoming more and more significant. Once a public opinion event breaks out and triggers a mass incident in Internet, it will greatly affect the social stability, such as the salt-buying panic caused by the catastrophic earthquake in Japan in 2011. Therefore, the evolution and prediction of public opinion are increasingly attracted researchers'interest, and have become the most prevalent topics at present. However, in most of the former research, the public opinion was evolved only in a single layer social network, ignoring the facts that the public opinion is not only propagated from mouth to mouth in real world but also in Internet, such as Microblog, WeChat and BBS etc.. Furthermore, the characteristics of public opinion in different regions may be diverse due to the varying of their compositions of netizens and their social structures. Hence, the regional evolution mechanism of public opinion in online and offline networks is a very important research topic. To this end, aiming at regional evolution mechanism of public opinion, this dissertation studied the propagation characteristics, the propagation mechanisms, the affecting factors and the propagation models of public opinion in online social network and in online and offline coupling network from the perspective of the compositions of netizens. The main research contents are as follows:(1) Statistically analyzing the region characteristics and the propagation characteristics of public opinion respectively, and then establishing a social vulnerability assessment model of regional public opinion. Empirical studies show that the social vulnerability of network public opinion in China has a distinct regional feature. More specifically, in order to correctly assess the regional social vulnerability, an assessment indicator system which contains twelve three-grade indicators was proposed from the following dimensionalities, i.e., the social-economic environment, the level of network information interaction and the government's emergency-dealing ability, then the social vulnerability of provincial-level regions in China in 2013 was empirical studied by combining entropy weight method and G1 weighting method. The network public opinion oriented regional social vulnerability assessment is a critical step for comprehending the network public opinion risk and enhancing the public opinion response capacity. The assessment results also help in providing a reference for regional public opinions pre-warning in macroscopic view.(2) Studying the evolution model of public opinion in Microblog network. Considering that only the retweet nodes influence their neighbor nodes during the information propagation and opinion evolution in Microblog network, based on social vulnerability theory, a regional public opinion dynamics model was established through integrating information propagation and opinion interaction pattern in Microblog network, and then on this basis the evolution characteristics of public opinion was discussed. More specifically, firstly, considering the fact that the time and energy netizens spent in Internet is limited, a Microblog network model was established by taking into account cognitive cost, anti-cognitive cost and the reciprocity mentality when Microblog users add friends. Secondly, considering the following phenomenon of social psychology in communication, i.e., the spiral of silence effect and the social reinforcement effect, two public opinion propagation models which meet these effects were presented respectively in Microblog Network. Thirdly, regarding the fact that nodes have multilayer influences in the information propagation paths and different nodes have different confidence levels in Microblog network due to the difference of the individual social attributes, a public opinion evolution model in which the effects of multilayer neighbor nodes were considered was built. Finally, viewing the mass media impact as external influence factor, the influence of mass media on public opinion evolution was studied through integrating the bounded confidence rule and the heterogeneous bounded confidence of node. It was found that the simulation result is consistent with the statistic data of some realistic examples.(3) Studying the regional public opinion evolution model in online and offline coupling network. Considering the fact that public opinion can propagate both in offline and online social networks, even in multilayer networks, the online and offline coupling network model was established, and then on this basis the public opinion dynamics was studied. Specifically, firstly, taking into account the characteristics of online and offline social networks, nodes was connected according to the similarity of node attributes, and then online and offline coupling network model was established by giving the one-to-one interdependency. On this basis, the structure of the coupling network and the propagation of public opinion in it were analyzed, and it was found by simulation that the propagation of rumor in online and offline coupling network is different from that in single layer network. After that, taking the promotion mechanism of propagation and the different compositions of node confident level into consideration, a two layered coupling network evolution model which contains the individual social attributes was proposed.In conclusion, this dissertation expands the theory and method of public opinion dynamics in Microblog network and multilayer coupling network. This research helps in understanding the regional evolution characteristics of public opinion, and the public opinion evolution models proposed in this dissertation have sufficient theoretical and practical significances for understanding the public opinion evolution, and the control and prediction of public opinion. |
PDF Full Text Request